agent-skills

This repository contains local agent skill definitions plus a small installer script that syncs them into the shared agent skills directory.

What’s here

• skills/ - source skill folders for this repo
• install-skills - syncs local skills into the shared skills directory

How it works

install-skills looks for subdirectories under skills/ that contain a SKILL.md file. Each matching skill is copied into:

• ~/.agents/skills/<skill-name> by default
• ~/.kiro/skills/<skill-name> when you pass --kiro

You can override the target location with:

• AGENTS_HOME
• AGENTS_SKILLS_DIR

The script also removes installed skills that no longer exist in the source tree, and it prunes skills that are on that target's ignore list, so the destination stays in sync.

Shared config

Some skill scripts read a shared JSON config file for non-interactive defaults and secrets:

• ~/.config/agent-skills/config.json

This file lives on your machine, outside the repo, so install-skills will not create or sync it.

Use sections to keep values organized by skill:

{
  "jenkins": {
    "url": "https://jenkins.example.com",
    "user": "your-user",
    "token": "your-api-token"
  },
  "jira": {
    "bearer_token": "your-bearer-token",
    "basic_user": "user@example.com",
    "basic_token": "your-api-token",
    "project": "ENG",
    "output_dir": "/tmp/jira-downloads"
  },
  "defaults": {
    "output_dir": "/tmp/shared-downloads"
  }
}

Scripts read their skill section first, then defaults, then top-level keys. That keeps one file usable across skills without forcing every script to share the same schema.

Usage

./install-skills

./install-skills --kiro

./install-skills --ignore github

Use --watch when you want to keep watching for changes and resync automatically.

--ignore <skill-name> is persistent for the selected target. If you ignore github on the default target, future default syncs will keep skipping it even if you omit the flag. The Kiro target keeps its own separate ignore list.

Current skills

• jira - Jira workflows via the local jira CLI, including issue search, editing, comments, sprint/epic operations, and attachment downloads with shared config
• github - GitHub workflows via the local gh CLI, including auth, repo targeting, issues, pull requests, workflows, releases, and API calls
• interrogate - alignment-first software development intake workflow that explores the codebase first and asks one question at a time until the request is clear
• jenkins - Jenkins workflows for branch-aware build lookup, failure and test inspection, console troubleshooting, and triggering parameterized builds from shared config
• neovim-plugin-dev - Neovim expertise and plugin development workflow with cached local help docs
• playwright - browser automation workflow for the Playwright coding-agent CLI, including snapshots, sessions, captures, and advanced debugging
• web-search - live web search and page fetch workflow via Exa's hosted MCP server using npx mcporter (web_search_exa, web_fetch_exa)

Archived skills

• spec-dev - archived staged spec-driven development workflow kept for reference only

Adding a new skill

1. Create a new folder under skills/.
2. Add a SKILL.md file to that folder.
3. Run ./install-skills to publish it to the shared skills directory, or ./install-skills --kiro to publish it to Kiro.
4. If needed, run ./install-skills --ignore <skill-name> to keep a skill excluded for that target going forward.